Artificial Intelligence Nanodegree

Convolutional Neural Networks

Project: Write an Algorithm for a Dog Identification App

---

In this notebook, some template code has already been provided for you, and you will need to implement additional functionality to successfully complete this project. You will not need to modify the included code beyond what is requested. Sections that begin with '(IMPLEMENTATION)' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!

Note: Once you have completed all of the code implementations, you need to finalize your work by exporting the iPython Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You can then export the notebook by using the menu above and navigating to \n", "File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.

In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question X' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.

Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. Markdown cells can be edited by double-clicking the cell to enter edit mode.

The rubric contains optional "Stand Out Suggestions" for enhancing the project beyond the minimum requirements. If you decide to pursue the "Stand Out Suggestions", you should include the code in this IPython notebook.

---

Why We're Here

In this notebook, you will make the first steps towards developing an algorithm that could be used as part of a mobile or web app. At the end of this project, your code will accept any user-supplied image as input. If a dog is detected in the image, it will provide an estimate of the dog's breed. If a human is detected, it will provide an estimate of the dog breed that is most resembling. The image below displays potential sample output of your finished project (... but we expect that each student's algorithm will behave differently!).

In this real-world setting, you will need to piece together a series of models to perform different tasks; for instance, the algorithm that detects humans in an image will be different from the CNN that infers dog breed. There are many points of possible failure, and no perfect algorithm exists. Your imperfect solution will nonetheless create a fun user experience!

The Road Ahead

We break the notebook into separate steps. Feel free to use the links below to navigate the notebook.

• Step 0: Import Datasets
• Step 1: Detect Humans
• Step 2: Detect Dogs
• Step 3: Create a CNN to Classify Dog Breeds (from Scratch)
• Step 4: Use a CNN to Classify Dog Breeds (using Transfer Learning)
• Step 5: Create a CNN to Classify Dog Breeds (using Transfer Learning)
• Step 6: Write your Algorithm
• Step 7: Test Your Algorithm

---

Step 0: Import Datasets

Import Dog Dataset

In the code cell below, we import a dataset of dog images. We populate a few variables through the use of the load_files function from the scikit-learn library:

• train_files, valid_files, test_files - numpy arrays containing file paths to images
• train_targets, valid_targets, test_targets - numpy arrays containing onehot-encoded classification labels
• dog_names - list of string-valued dog breed names for translating labels

from sklearn.datasets import load_files       
from keras.utils import np_utils
import numpy as np
from glob import glob

# define function to load train, test, and validation datasets
def load_dataset(path):
    data = load_files(path)
    dog_files = np.array(data['filenames'])
    dog_targets = np_utils.to_categorical(np.array(data['target']), 133)
    return dog_files, dog_targets

# load train, test, and validation datasets
train_files, train_targets = load_dataset('dogImages/train')
valid_files, valid_targets = load_dataset('dogImages/valid')
test_files, test_targets = load_dataset('dogImages/test')

# load list of dog names
dog_names = [item[20:-1] for item in sorted(glob("dogImages/train/*/"))]

# print statistics about the dataset
print('There are %d total dog categories.' % len(dog_names))
print('There are %s total dog images.\n' % len(np.hstack([train_files, valid_files, test_files])))
print('There are %d training dog images.' % len(train_files))
print('There are %d validation dog images.' % len(valid_files))
print('There are %d test dog images.'% len(test_files))

Using TensorFlow backend.

There are 133 total dog categories.
There are 8351 total dog images.

There are 6680 training dog images.
There are 835 validation dog images.
There are 836 test dog images.

Import Human Dataset

In the code cell below, we import a dataset of human images, where the file paths are stored in the numpy array human_files.

import random
random.seed(8675309)

# load filenames in shuffled human dataset
human_files = np.array(glob("lfw/*/*"))
random.shuffle(human_files)

# print statistics about the dataset
print('There are %d total human images.' % len(human_files))

There are 13233 total human images.

---

Step 1: Detect Humans

We use OpenCV's implementation of Haar feature-based cascade classifiers to detect human faces in images. OpenCV provides many pre-trained face detectors, stored as XML files on github. We have downloaded one of these detectors and stored it in the haarcascades directory.

In the next code cell, we demonstrate how to use this detector to find human faces in a sample image.

import cv2                
import matplotlib.pyplot as plt                        
%matplotlib inline                               

# extract pre-trained face detector
face_cascade = cv2.CascadeClassifier('haarcascades/haarcascade_frontalface_alt.xml')

# load color (BGR) image
img = cv2.imread(human_files[3])
# convert BGR image to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# find faces in image
faces = face_cascade.detectMultiScale(gray)

# print number of faces detected in the image
print('Number of faces detected:', len(faces))

# get bounding box for each detected face
for (x,y,w,h) in faces:
    # add bounding box to color image
    cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)
    
# convert BGR image to RGB for plotting
cv_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

# display the image, along with bounding box
plt.imshow(cv_rgb)
plt.show()

Number of faces detected: 1

Before using any of the face detectors, it is standard procedure to convert the images to grayscale. The detectMultiScale function executes the classifier stored in face_cascade and takes the grayscale image as a parameter.

In the above code, faces is a numpy array of detected faces, where each row corresponds to a detected face. Each detected face is a 1D array with four entries that specifies the bounding box of the detected face. The first two entries in the array (extracted in the above code as x and y) specify the horizontal and vertical positions of the top left corner of the bounding box. The last two entries in the array (extracted here as w and h) specify the width and height of the box.

Write a Human Face Detector

We can use this procedure to write a function that returns True if a human face is detected in an image and False otherwise. This function, aptly named face_detector, takes a string-valued file path to an image as input and appears in the code block below.

# returns "True" if face is detected in image stored at img_path
def face_detector(img_path):
    img = cv2.imread(img_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray)
    return len(faces) > 0

(IMPLEMENTATION) Assess the Human Face Detector

Question 1: Use the code cell below to test the performance of the face_detector function.

• What percentage of the first 100 images in human_files have a detected human face?
• What percentage of the first 100 images in dog_files have a detected human face?

Ideally, we would like 100% of human images with a detected face and 0% of dog images with a detected face. You will see that our algorithm falls short of this goal, but still gives acceptable performance. We extract the file paths for the first 100 images from each of the datasets and store them in the numpy arrays human_files_short and dog_files_short.

Answer: All of the images with human faces were classified correctly and human faces were detected on every image (100/100, 100%). From 100 dogs images, haar classifier detected human faces on 11 images (11/100, 89% correctness). This is acceptable performance.

human_files_short = human_files[:100]
dog_files_short = train_files[:100]
# Do NOT modify the code above this line.

## TODO: Test the performance of the face_detector algorithm 
## on the images in human_files_short and dog_files_short.
human_faces_detected = 0
dog_faces_detected = 0
for i in range(100):
    human_faces_detected += face_detector(human_files_short[i])
    dog_faces_detected += face_detector(dog_files_short[i])

print("From 100 images with human faces detected faces: ", human_faces_detected)
print("From 100 images with dogs detected human faces: ", dog_faces_detected)

From 100 images with human faces detected faces:  100
From 100 images with dogs detected human faces:  11

Question 2: This algorithmic choice necessitates that we communicate to the user that we accept human images only when they provide a clear view of a face (otherwise, we risk having unneccessarily frustrated users!). In your opinion, is this a reasonable expectation to pose on the user? If not, can you think of a way to detect humans in images that does not necessitate an image with a clearly presented face?

Answer: I think that Haar classifier is one of the best choices when finding something "familiar" on the image and labelling it. Alternatively, we can train a separate CNN for detecting humans, but to be able to detect a human somewhere on a big image with lots of other objects, we need to use something like sliding windows technique.

We suggest the face detector from OpenCV as a potential way to detect human images in your algorithm, but you are free to explore other approaches, especially approaches that make use of deep learning :). Please use the code cell below to design and test your own face detection algorithm. If you decide to pursue this optional task, report performance on each of the datasets.

## (Optional) TODO: Report the performance of another  
## face detection algorithm on the LFW dataset
### Feel free to use as many code cells as needed.

---

Step 2: Detect Dogs

In this section, we use a pre-trained ResNet-50 model to detect dogs in images. Our first line of code downloads the ResNet-50 model, along with weights that have been trained on ImageNet, a very large, very popular dataset used for image classification and other vision tasks. ImageNet contains over 10 million URLs, each linking to an image containing an object from one of 1000 categories. Given an image, this pre-trained ResNet-50 model returns a prediction (derived from the available categories in ImageNet) for the object that is contained in the image.

from keras.applications.resnet50 import ResNet50

# define ResNet50 model
ResNet50_model = ResNet50(weights='imagenet')

Downloading data from https://github.com/fchollet/deep-learning-models/releases/download/v0.2/resnet50_weights_tf_dim_ordering_tf_kernels.h5

Pre-process the Data

When using TensorFlow as backend, Keras CNNs require a 4D array (which we'll also refer to as a 4D tensor) as input, with shape

$$ (\text{nb_samples}, \text{rows}, \text{columns}, \text{channels}), $$

where nb_samples corresponds to the total number of images (or samples), and rows, columns, and channels correspond to the number of rows, columns, and channels for each image, respectively.

The path_to_tensor function below takes a string-valued file path to a color image as input and returns a 4D tensor suitable for supplying to a Keras CNN. The function first loads the image and resizes it to a square image that is $224 \times 224$ pixels. Next, the image is converted to an array, which is then resized to a 4D tensor. In this case, since we are working with color images, each image has three channels. Likewise, since we are processing a single image (or sample), the returned tensor will always have shape

$$ (1, 224, 224, 3). $$

The paths_to_tensor function takes a numpy array of string-valued image paths as input and returns a 4D tensor with shape

$$ (\text{nb_samples}, 224, 224, 3). $$

Here, nb_samples is the number of samples, or number of images, in the supplied array of image paths. It is best to think of nb_samples as the number of 3D tensors (where each 3D tensor corresponds to a different image) in your dataset!

from keras.preprocessing import image                  
from tqdm import tqdm

def path_to_tensor(img_path):
    # loads RGB image as PIL.Image.Image type
    img = image.load_img(img_path, target_size=(224, 224))
    # convert PIL.Image.Image type to 3D tensor with shape (224, 224, 3)
    x = image.img_to_array(img)
    # convert 3D tensor to 4D tensor with shape (1, 224, 224, 3) and return 4D tensor
    return np.expand_dims(x, axis=0)

def paths_to_tensor(img_paths):
    list_of_tensors = [path_to_tensor(img_path) for img_path in tqdm(img_paths)]
    return np.vstack(list_of_tensors)

Making Predictions with ResNet-50

Getting the 4D tensor ready for ResNet-50, and for any other pre-trained model in Keras, requires some additional processing. First, the RGB image is converted to BGR by reordering the channels. All pre-trained models have the additional normalization step that the mean pixel (expressed in RGB as $[103.939, 116.779, 123.68]$ and calculated from all pixels in all images in ImageNet) must be subtracted from every pixel in each image. This is implemented in the imported function preprocess_input. If you're curious, you can check the code for preprocess_input here.

Now that we have a way to format our image for supplying to ResNet-50, we are now ready to use the model to extract the predictions. This is accomplished with the predict method, which returns an array whose $i$-th entry is the model's predicted probability that the image belongs to the $i$-th ImageNet category. This is implemented in the ResNet50_predict_labels function below.

By taking the argmax of the predicted probability vector, we obtain an integer corresponding to the model's predicted object class, which we can identify with an object category through the use of this dictionary.

from keras.applications.resnet50 import preprocess_input, decode_predictions

def ResNet50_predict_labels(img_path):
    # returns prediction vector for image located at img_path
    img = preprocess_input(path_to_tensor(img_path))
    return np.argmax(ResNet50_model.predict(img))

Write a Dog Detector

While looking at the dictionary, you will notice that the categories corresponding to dogs appear in an uninterrupted sequence and correspond to dictionary keys 151-268, inclusive, to include all categories from 'Chihuahua' to 'Mexican hairless'. Thus, in order to check to see if an image is predicted to contain a dog by the pre-trained ResNet-50 model, we need only check if the ResNet50_predict_labels function above returns a value between 151 and 268 (inclusive).

We use these ideas to complete the dog_detector function below, which returns True if a dog is detected in an image (and False if not).

### returns "True" if a dog is detected in the image stored at img_path
def dog_detector(img_path):
    prediction = ResNet50_predict_labels(img_path)
    return ((prediction <= 268) & (prediction >= 151)) 

(IMPLEMENTATION) Assess the Dog Detector

Question 3: Use the code cell below to test the performance of your dog_detector function.

• What percentage of the images in human_files_short have a detected dog?
• What percentage of the images in dog_files_short have a detected dog?

Answer: Detector worked perfectly. It correctly detected dogs on dogs images with 100% accuracy and haven't detected any dogs on human images (also 100% accuracy)

### TODO: Test the performance of the dog_detector function
### on the images in human_files_short and dog_files_short.
human_imgs_dog_detected = 0
dog_imgs_dog_detected = 0
for i in range(100):
    human_imgs_dog_detected += dog_detector(human_files_short[i])
    dog_imgs_dog_detected += dog_detector(dog_files_short[i])

print("From 100 images with human faces detected dogs: ", human_imgs_dog_detected)
print("From 100 images with dogs detected dogs: ", dog_imgs_dog_detected)

From 100 images with human faces detected dogs:  0
From 100 images with dogs detected dogs:  100

---

Step 3: Create a CNN to Classify Dog Breeds (from Scratch)

Now that we have functions for detecting humans and dogs in images, we need a way to predict breed from images. In this step, you will create a CNN that classifies dog breeds. You must create your CNN from scratch (so, you can't use transfer learning yet!), and you must attain a test accuracy of at least 1%. In Step 5 of this notebook, you will have the opportunity to use transfer learning to create a CNN that attains greatly improved accuracy.

Be careful with adding too many trainable layers! More parameters means longer training, which means you are more likely to need a GPU to accelerate the training process. Thankfully, Keras provides a handy estimate of the time that each epoch is likely to take; you can extrapolate this estimate to figure out how long it will take for your algorithm to train.

We mention that the task of assigning breed to dogs from images is considered exceptionally challenging. To see why, consider that even a human would have great difficulty in distinguishing between a Brittany and a Welsh Springer Spaniel.

Brittany	Welsh Springer Spaniel

It is not difficult to find other dog breed pairs with minimal inter-class variation (for instance, Curly-Coated Retrievers and American Water Spaniels).

Curly-Coated Retriever	American Water Spaniel

Likewise, recall that labradors come in yellow, chocolate, and black. Your vision-based algorithm will have to conquer this high intra-class variation to determine how to classify all of these different shades as the same breed.

Yellow Labrador	Chocolate Labrador	Black Labrador

We also mention that random chance presents an exceptionally low bar: setting aside the fact that the classes are slightly imabalanced, a random guess will provide a correct answer roughly 1 in 133 times, which corresponds to an accuracy of less than 1%.

Remember that the practice is far ahead of the theory in deep learning. Experiment with many different architectures, and trust your intuition. And, of course, have fun!

Pre-process the Data

We rescale the images by dividing every pixel in every image by 255.

from PIL import ImageFile                            
ImageFile.LOAD_TRUNCATED_IMAGES = True                 

# pre-process the data for Keras
train_tensors = paths_to_tensor(train_files).astype('float32')/255
valid_tensors = paths_to_tensor(valid_files).astype('float32')/255
test_tensors = paths_to_tensor(test_files).astype('float32')/255

100%|██████████| 6680/6680 [01:31<00:00, 73.28it/s] 
100%|██████████| 835/835 [00:10<00:00, 81.81it/s] 
100%|██████████| 836/836 [00:10<00:00, 82.18it/s] 

(IMPLEMENTATION) Model Architecture

Create a CNN to classify dog breed. At the end of your code cell block, summarize the layers of your model by executing the line:

    model.summary()

We have imported some Python modules to get you started, but feel free to import as many modules as you need. If you end up getting stuck, here's a hint that specifies a model that trains relatively fast on CPU and attains >1% test accuracy in 5 epochs:

Question 4: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. If you chose to use the hinted architecture above, describe why you think that CNN architecture should work well for the image classification task.

Answer: I have started with the hinted architecture and then slightly modified it via experimenting with different layers.

• First of all, I created two convolutional layers with 16 and 32 filters without pooling layers between them not to lose important features detected on a low level. As input image has pretty big dimensions (224x224), I've chosen kernal_size of 5 and strides of 2 for the first layer to reduce dimensions.
• After that I included max pooling layer to reduce number of trainable parameters
• I followed by including small dropout to prevent overfitting
• Then I included final convolution layer with 64 filters and 'same' padding to build feature map of top level features
• Another max pooling layer for parameters number reducing and small dropout for overfit prevention
• GAP to narrow down number of parameters
• Fully connected layer with final results

I have also experimented with the final part of the model, adding two and three fully-connected layers and replacing GAP with Dense layer, but it hasn't given significant improvement (even made results worse sometimes) and took more time to train (at maximum, the number of params was about 132 millions) The result model trained for 5 epochs and gave nearly 3% accuracy on the test set.

from keras.layers import Conv2D, MaxPooling2D, GlobalAveragePooling2D
from keras.layers import Dropout, Flatten, Dense
from keras.models import Sequential

model = Sequential()

### TODO: Define your architecture.
model.add(Conv2D(filters=16, kernel_size=5, strides=2, padding='valid', activation='relu', input_shape=(224, 224, 3)))
model.add(Conv2D(filters=32, kernel_size=3, strides=2, padding='valid', activation='relu'))
model.add(MaxPooling2D(pool_size=2, strides=1))
model.add(Dropout(.2))
model.add(Conv2D(filters=64, kernel_size=3, strides=2, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2, strides=1))
model.add(Dropout(.2))
model.add(GlobalAveragePooling2D())
model.add(Dense(133, activation='softmax'))

model.summary()

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d_22 (Conv2D)           (None, 110, 110, 16)      1216      
_________________________________________________________________
conv2d_23 (Conv2D)           (None, 54, 54, 32)        4640      
_________________________________________________________________
max_pooling2d_16 (MaxPooling (None, 53, 53, 32)        0         
_________________________________________________________________
dropout_17 (Dropout)         (None, 53, 53, 32)        0         
_________________________________________________________________
conv2d_24 (Conv2D)           (None, 27, 27, 64)        18496     
_________________________________________________________________
max_pooling2d_17 (MaxPooling (None, 26, 26, 64)        0         
_________________________________________________________________
dropout_18 (Dropout)         (None, 26, 26, 64)        0         
_________________________________________________________________
global_average_pooling2d_4 ( (None, 64)                0         
_________________________________________________________________
dense_9 (Dense)              (None, 133)               8645      
=================================================================
Total params: 32,997.0
Trainable params: 32,997.0
Non-trainable params: 0.0
_________________________________________________________________

Compile the Model

model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])

(IMPLEMENTATION) Train the Model

Train your model in the code cell below. Use model checkpointing to save the model that attains the best validation loss.

You are welcome to augment the training data, but this is not a requirement.

from keras.callbacks import ModelCheckpoint  

### TODO: specify the number of epochs that you would like to use to train the model.

epochs = 5

### Do NOT modify the code below this line.

checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.from_scratch.hdf5', 
                               verbose=1, save_best_only=True)

model.fit(train_tensors, train_targets, 
          validation_data=(valid_tensors, valid_targets),
          epochs=epochs, batch_size=20, callbacks=[checkpointer], verbose=1)

Train on 6680 samples, validate on 835 samples
Epoch 1/5
6660/6680 [============================>.] - ETA: 0s - loss: 4.8694 - acc: 0.0102   - ETA: 35s - loss: 4.8867 - acc: 0.0085Epoch 00000: val_loss improved from inf to 4.82233, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 69s - loss: 4.8690 - acc: 0.0102 - val_loss: 4.8223 - val_acc: 0.0192
Epoch 2/5
6660/6680 [============================>.] - ETA: 0s - loss: 4.8093 - acc: 0.0146     Epoch 00001: val_loss improved from 4.82233 to 4.78546, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 69s - loss: 4.8090 - acc: 0.0145 - val_loss: 4.7855 - val_acc: 0.0216
Epoch 3/5
6660/6680 [============================>.] - ETA: 0s - loss: 4.7649 - acc: 0.0180      - ETA: 33s - loss: 4.7690 - acc: 0.0203 - ETA: 17s - loss: 4.7660 - acc: 0.0185 - ETA: 10s - loss: 4.7648 - acc: 0.0186Epoch 00002: val_loss improved from 4.78546 to 4.75056, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 69s - loss: 4.7646 - acc: 0.0183 - val_loss: 4.7506 - val_acc: 0.0299
Epoch 4/5
6660/6680 [============================>.] - ETA: 0s - loss: 4.7243 - acc: 0.0225      - ETA: 11s - loss: 4.7233 - acc: 0.0224Epoch 00003: val_loss improved from 4.75056 to 4.70375, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 69s - loss: 4.7246 - acc: 0.0225 - val_loss: 4.7037 - val_acc: 0.0311
Epoch 5/5
6660/6680 [============================>.] - ETA: 0s - loss: 4.6801 - acc: 0.0296      - ETA: 52s - loss: 4.6547 - acc: 0.0254 - ETA: 27s - loss: 4.6785 - acc: 0.0276 - ETA: 8s - loss: 4.6801 - acc: 0.0281Epoch 00004: val_loss improved from 4.70375 to 4.69173, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 69s - loss: 4.6803 - acc: 0.0295 - val_loss: 4.6917 - val_acc: 0.0275

<keras.callbacks.History at 0x12c1a5f60>

Load the Model with the Best Validation Loss

model.load_weights('saved_models/weights.best.from_scratch.hdf5')

Test the Model

Try out your model on the test dataset of dog images. Ensure that your test accuracy is greater than 1%.

# get index of predicted dog breed for each image in test set
dog_breed_predictions = [np.argmax(model.predict(np.expand_dims(tensor, axis=0))) for tensor in test_tensors]

# report test accuracy
test_accuracy = 100*np.sum(np.array(dog_breed_predictions)==np.argmax(test_targets, axis=1))/len(dog_breed_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)

Test accuracy: 2.9904%

---

Step 4: Use a CNN to Classify Dog Breeds

To reduce training time without sacrificing accuracy, we show you how to train a CNN using transfer learning. In the following step, you will get a chance to use transfer learning to train your own CNN.

Obtain Bottleneck Features

bottleneck_features = np.load('bottleneck_features/DogVGG16Data.npz')
train_VGG16 = bottleneck_features['train']
valid_VGG16 = bottleneck_features['valid']
test_VGG16 = bottleneck_features['test']

Model Architecture

The model uses the the pre-trained VGG-16 model as a fixed feature extractor, where the last convolutional output of VGG-16 is fed as input to our model. We only add a global average pooling layer and a fully connected layer, where the latter contains one node for each dog category and is equipped with a softmax.

VGG16_model = Sequential()
VGG16_model.add(GlobalAveragePooling2D(input_shape=train_VGG16.shape[1:]))
VGG16_model.add(Dense(133, activation='softmax'))

VGG16_model.summary()

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
global_average_pooling2d_5 ( (None, 512)               0         
_________________________________________________________________
dense_10 (Dense)             (None, 133)               68229     
=================================================================
Total params: 68,229.0
Trainable params: 68,229.0
Non-trainable params: 0.0
_________________________________________________________________

Compile the Model

VGG16_model.compile(loss='categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])

Train the Model

checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.VGG16.hdf5', 
                               verbose=1, save_best_only=True)

VGG16_model.fit(train_VGG16, train_targets, 
          validation_data=(valid_VGG16, valid_targets),
          epochs=20, batch_size=20, callbacks=[checkpointer], verbose=1)

Train on 6680 samples, validate on 835 samples
Epoch 1/20
6220/6680 [==========================>...] - ETA: 0s - loss: 12.5071 - acc: 0.1159     Epoch 00000: val_loss improved from inf to 10.93984, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 12.3571 - acc: 0.1250 - val_loss: 10.9398 - val_acc: 0.2168
Epoch 2/20
6620/6680 [============================>.] - ETA: 0s - loss: 10.3849 - acc: 0.2773Epoch 00001: val_loss improved from 10.93984 to 10.36179, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 10.3782 - acc: 0.2778 - val_loss: 10.3618 - val_acc: 0.2743
Epoch 3/20
6640/6680 [============================>.] - ETA: 0s - loss: 10.0323 - acc: 0.3277Epoch 00002: val_loss improved from 10.36179 to 10.24237, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 10.0359 - acc: 0.3274 - val_loss: 10.2424 - val_acc: 0.2982
Epoch 4/20
6560/6680 [============================>.] - ETA: 0s - loss: 9.9158 - acc: 0.3520 Epoch 00003: val_loss improved from 10.24237 to 10.01913, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.8831 - acc: 0.3540 - val_loss: 10.0191 - val_acc: 0.3210
Epoch 5/20
6660/6680 [============================>.] - ETA: 0s - loss: 9.7026 - acc: 0.3706 Epoch 00004: val_loss improved from 10.01913 to 9.99185, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.6968 - acc: 0.3710 - val_loss: 9.9918 - val_acc: 0.3210
Epoch 6/20
6180/6680 [==========================>...] - ETA: 0s - loss: 9.5253 - acc: 0.3833Epoch 00005: val_loss improved from 9.99185 to 9.82506, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.5205 - acc: 0.3837 - val_loss: 9.8251 - val_acc: 0.3401
Epoch 7/20
6180/6680 [==========================>...] - ETA: 0s - loss: 9.3019 - acc: 0.3990 Epoch 00006: val_loss improved from 9.82506 to 9.61433, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.3182 - acc: 0.3985 - val_loss: 9.6143 - val_acc: 0.3485
Epoch 8/20
6620/6680 [============================>.] - ETA: 0s - loss: 9.1815 - acc: 0.4136Epoch 00007: val_loss improved from 9.61433 to 9.50605, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.1886 - acc: 0.4132 - val_loss: 9.5060 - val_acc: 0.3545
Epoch 9/20
6160/6680 [==========================>...] - ETA: 0s - loss: 9.1066 - acc: 0.4213Epoch 00008: val_loss improved from 9.50605 to 9.39569, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.1281 - acc: 0.4198 - val_loss: 9.3957 - val_acc: 0.3725
Epoch 10/20
6320/6680 [===========================>..] - ETA: 0s - loss: 9.0963 - acc: 0.4220Epoch 00009: val_loss improved from 9.39569 to 9.36214, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.0412 - acc: 0.4256 - val_loss: 9.3621 - val_acc: 0.3737
Epoch 11/20
6420/6680 [===========================>..] - ETA: 0s - loss: 8.9901 - acc: 0.4326 Epoch 00010: val_loss improved from 9.36214 to 9.32597, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.9919 - acc: 0.4325 - val_loss: 9.3260 - val_acc: 0.3760
Epoch 12/20
6180/6680 [==========================>...] - ETA: 0s - loss: 8.9937 - acc: 0.4330 Epoch 00011: val_loss improved from 9.32597 to 9.22248, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.9344 - acc: 0.4355 - val_loss: 9.2225 - val_acc: 0.3665
Epoch 13/20
6180/6680 [==========================>...] - ETA: 0s - loss: 8.7792 - acc: 0.4393Epoch 00012: val_loss improved from 9.22248 to 9.07452, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.7556 - acc: 0.4407 - val_loss: 9.0745 - val_acc: 0.3820
Epoch 14/20
6660/6680 [============================>.] - ETA: 0s - loss: 8.6643 - acc: 0.4526Epoch 00013: val_loss improved from 9.07452 to 9.03993, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.6697 - acc: 0.4522 - val_loss: 9.0399 - val_acc: 0.3856
Epoch 15/20
6180/6680 [==========================>...] - ETA: 0s - loss: 8.5430 - acc: 0.4552 Epoch 00014: val_loss improved from 9.03993 to 9.00548, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.5430 - acc: 0.4545 - val_loss: 9.0055 - val_acc: 0.3784
Epoch 16/20
6500/6680 [============================>.] - ETA: 0s - loss: 8.2141 - acc: 0.4692Epoch 00015: val_loss improved from 9.00548 to 8.71335, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.2357 - acc: 0.4678 - val_loss: 8.7134 - val_acc: 0.3940
Epoch 17/20
6200/6680 [==========================>...] - ETA: 0s - loss: 8.0883 - acc: 0.4834Epoch 00016: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 8.1009 - acc: 0.4831 - val_loss: 8.7206 - val_acc: 0.3856
Epoch 18/20
6180/6680 [==========================>...] - ETA: 0s - loss: 8.0077 - acc: 0.4890Epoch 00017: val_loss improved from 8.71335 to 8.50625, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 7.9951 - acc: 0.4901 - val_loss: 8.5063 - val_acc: 0.4120
Epoch 19/20
6660/6680 [============================>.] - ETA: 0s - loss: 7.9351 - acc: 0.4994Epoch 00018: val_loss improved from 8.50625 to 8.49027, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 7.9311 - acc: 0.4996 - val_loss: 8.4903 - val_acc: 0.4108
Epoch 20/20
6180/6680 [==========================>...] - ETA: 0s - loss: 7.8245 - acc: 0.5011Epoch 00019: val_loss improved from 8.49027 to 8.36073, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 7.8076 - acc: 0.5027 - val_loss: 8.3607 - val_acc: 0.4168

<keras.callbacks.History at 0x12cc55e48>

Load the Model with the Best Validation Loss

VGG16_model.load_weights('saved_models/weights.best.VGG16.hdf5')

Test the Model

Now, we can use the CNN to test how well it identifies breed within our test dataset of dog images. We print the test accuracy below.

# get index of predicted dog breed for each image in test set
VGG16_predictions = [np.argmax(VGG16_model.predict(np.expand_dims(feature, axis=0))) for feature in test_VGG16]

# report test accuracy
test_accuracy = 100*np.sum(np.array(VGG16_predictions)==np.argmax(test_targets, axis=1))/len(VGG16_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)

Test accuracy: 42.4641%

Predict Dog Breed with the Model

from extract_bottleneck_features import *

def VGG16_predict_breed(img_path):
    # extract bottleneck features
    bottleneck_feature = extract_VGG16(path_to_tensor(img_path))
    # obtain predicted vector
    predicted_vector = VGG16_model.predict(bottleneck_feature)
    # return dog breed that is predicted by the model
    return dog_names[np.argmax(predicted_vector)]

---

Step 5: Create a CNN to Classify Dog Breeds (using Transfer Learning)

You will now use transfer learning to create a CNN that can identify dog breed from images. Your CNN must attain at least 60% accuracy on the test set.

In Step 4, we used transfer learning to create a CNN using VGG-16 bottleneck features. In this section, you must use the bottleneck features from a different pre-trained model. To make things easier for you, we have pre-computed the features for all of the networks that are currently available in Keras:

• VGG-19 bottleneck features
• ResNet-50 bottleneck features
• Inception bottleneck features
• Xception bottleneck features

The files are encoded as such:

Dog{network}Data.npz

where {network}, in the above filename, can be one of VGG19, Resnet50, InceptionV3, or Xception. Pick one of the above architectures, download the corresponding bottleneck features, and store the downloaded file in the bottleneck_features/ folder in the repository.

(IMPLEMENTATION) Obtain Bottleneck Features

In the code block below, extract the bottleneck features corresponding to the train, test, and validation sets by running the following:

bottleneck_features = np.load('bottleneck_features/Dog{network}Data.npz')
train_{network} = bottleneck_features['train']
valid_{network} = bottleneck_features['valid']
test_{network} = bottleneck_features['test']

### TODO: Obtain bottleneck features from another pre-trained CNN.
bottleneck_features = np.load('bottleneck_features/DogResnet50Data.npz')
train_resnet50 = bottleneck_features['train']
valid_resnet50 = bottleneck_features['valid']
test_resnet50 = bottleneck_features['test']

(IMPLEMENTATION) Model Architecture

Create a CNN to classify dog breed. At the end of your code cell block, summarize the layers of your model by executing the line:

    <your model's name>.summary()

Question 5: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. Describe why you think the architecture is suitable for the current problem.

Answer: In the provided above example a GAP layer was used. I've decided to replace it with a simple Flatten layer, as we are training only one layer which should not take very long. Moreover, such approach would lead to not losing important information (as global pooling would have narrowed down the feature map). Last layer is fully connected layer with exact number of available breeds for correct classification. Such approach reported 80% on the test set. Pretty acceptable.

### TODO: Define your architecture.
resnet50_model = Sequential()
resnet50_model.add(Flatten(input_shape=train_resnet50.shape[1:]))
resnet50_model.add(Dense(133, activation='softmax'))

resnet50_model.summary()

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
flatten_8 (Flatten)          (None, 2048)              0         
_________________________________________________________________
dense_17 (Dense)             (None, 133)               272517    
=================================================================
Total params: 272,517.0
Trainable params: 272,517.0
Non-trainable params: 0.0
_________________________________________________________________

(IMPLEMENTATION) Compile the Model

### TODO: Compile the model.
resnet50_model.compile(loss='categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])

(IMPLEMENTATION) Train the Model

Train your model in the code cell below. Use model checkpointing to save the model that attains the best validation loss.

You are welcome to augment the training data, but this is not a requirement.

### TODO: Train the model.
checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.ResNet50.hdf5', 
                               verbose=1, save_best_only=True)

resnet50_model.fit(train_resnet50, train_targets, 
          validation_data=(valid_resnet50, valid_targets),
          epochs=25, batch_size=20, callbacks=[checkpointer], verbose=1)

Train on 6680 samples, validate on 835 samples
Epoch 1/25
6480/6680 [============================>.] - ETA: 0s - loss: 1.6304 - acc: 0.5977      Epoch 00000: val_loss improved from inf to 0.82132, saving model to saved_models/weights.best.ResNet50.hdf5
6680/6680 [==============================] - 1s - loss: 1.6068 - acc: 0.6021 - val_loss: 0.8213 - val_acc: 0.7437
Epoch 2/25
6660/6680 [============================>.] - ETA: 0s - loss: 0.4351 - acc: 0.8605Epoch 00001: val_loss improved from 0.82132 to 0.70151, saving model to saved_models/weights.best.ResNet50.hdf5
6680/6680 [==============================] - 0s - loss: 0.4343 - acc: 0.8606 - val_loss: 0.7015 - val_acc: 0.7892
Epoch 3/25
6660/6680 [============================>.] - ETA: 0s - loss: 0.2570 - acc: 0.9180Epoch 00002: val_loss improved from 0.70151 to 0.66807, saving model to saved_models/weights.best.ResNet50.hdf5
6680/6680 [==============================] - 0s - loss: 0.2577 - acc: 0.9181 - val_loss: 0.6681 - val_acc: 0.7976
Epoch 4/25
6600/6680 [============================>.] - ETA: 0s - loss: 0.1794 - acc: 0.9448Epoch 00003: val_loss improved from 0.66807 to 0.64714, saving model to saved_models/weights.best.ResNet50.hdf5
6680/6680 [==============================] - 0s - loss: 0.1799 - acc: 0.9445 - val_loss: 0.6471 - val_acc: 0.8120
Epoch 5/25
6640/6680 [============================>.] - ETA: 0s - loss: 0.1215 - acc: 0.9622Epoch 00004: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.1211 - acc: 0.9624 - val_loss: 0.6508 - val_acc: 0.8144
Epoch 6/25
6520/6680 [============================>.] - ETA: 0s - loss: 0.0855 - acc: 0.9725Epoch 00005: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0852 - acc: 0.9720 - val_loss: 0.7030 - val_acc: 0.8072
Epoch 7/25
6620/6680 [============================>.] - ETA: 0s - loss: 0.0620 - acc: 0.9820Epoch 00006: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0620 - acc: 0.9819 - val_loss: 0.7008 - val_acc: 0.8180
Epoch 8/25
6320/6680 [===========================>..] - ETA: 0s - loss: 0.0455 - acc: 0.9878Epoch 00007: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0461 - acc: 0.9870 - val_loss: 0.7248 - val_acc: 0.8120
Epoch 9/25
6360/6680 [===========================>..] - ETA: 0s - loss: 0.0358 - acc: 0.9888Epoch 00008: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0356 - acc: 0.9889 - val_loss: 0.7488 - val_acc: 0.8084
Epoch 10/25
6460/6680 [============================>.] - ETA: 0s - loss: 0.0248 - acc: 0.9935Epoch 00009: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0256 - acc: 0.9934 - val_loss: 0.7501 - val_acc: 0.8299
Epoch 11/25
6580/6680 [============================>.] - ETA: 0s - loss: 0.0185 - acc: 0.9960Epoch 00010: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0187 - acc: 0.9958 - val_loss: 0.7795 - val_acc: 0.8168
Epoch 12/25
6600/6680 [============================>.] - ETA: 0s - loss: 0.0162 - acc: 0.9955Epoch 00011: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0162 - acc: 0.9954 - val_loss: 0.7909 - val_acc: 0.8180
Epoch 13/25
6620/6680 [============================>.] - ETA: 0s - loss: 0.0123 - acc: 0.9971Epoch 00012: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0122 - acc: 0.9972 - val_loss: 0.8106 - val_acc: 0.8240
Epoch 14/25
6620/6680 [============================>.] - ETA: 0s - loss: 0.0125 - acc: 0.9967Epoch 00013: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0133 - acc: 0.9966 - val_loss: 0.8140 - val_acc: 0.8192
Epoch 15/25
6600/6680 [============================>.] - ETA: 0s - loss: 0.0100 - acc: 0.9977Epoch 00014: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0101 - acc: 0.9976 - val_loss: 0.8290 - val_acc: 0.8263
Epoch 16/25
6600/6680 [============================>.] - ETA: 0s - loss: 0.0079 - acc: 0.9982Epoch 00015: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0079 - acc: 0.9982 - val_loss: 0.8187 - val_acc: 0.8204
Epoch 17/25
6580/6680 [============================>.] - ETA: 0s - loss: 0.0082 - acc: 0.9980    Epoch 00016: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0081 - acc: 0.9981 - val_loss: 0.8511 - val_acc: 0.8192
Epoch 18/25
6600/6680 [============================>.] - ETA: 0s - loss: 0.0071 - acc: 0.9980    Epoch 00017: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0071 - acc: 0.9981 - val_loss: 0.9001 - val_acc: 0.8275
Epoch 19/25
6600/6680 [============================>.] - ETA: 0s - loss: 0.0057 - acc: 0.9985Epoch 00018: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0058 - acc: 0.9985 - val_loss: 0.9040 - val_acc: 0.8287
Epoch 20/25
6460/6680 [============================>.] - ETA: 0s - loss: 0.0060 - acc: 0.9981    Epoch 00019: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0059 - acc: 0.9982 - val_loss: 0.9495 - val_acc: 0.8240
Epoch 21/25
6660/6680 [============================>.] - ETA: 0s - loss: 0.0059 - acc: 0.9985    Epoch 00020: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0059 - acc: 0.9985 - val_loss: 0.9298 - val_acc: 0.8323
Epoch 22/25
6640/6680 [============================>.] - ETA: 0s - loss: 0.0052 - acc: 0.9988    Epoch 00021: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0051 - acc: 0.9988 - val_loss: 0.9118 - val_acc: 0.8204
Epoch 23/25
6260/6680 [===========================>..] - ETA: 0s - loss: 0.0059 - acc: 0.9989    Epoch 00022: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0056 - acc: 0.9990 - val_loss: 0.9665 - val_acc: 0.8263
Epoch 24/25
6540/6680 [============================>.] - ETA: 0s - loss: 0.0061 - acc: 0.9986    Epoch 00023: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0060 - acc: 0.9987 - val_loss: 1.0137 - val_acc: 0.8299
Epoch 25/25
6480/6680 [============================>.] - ETA: 0s - loss: 0.0051 - acc: 0.9991    Epoch 00024: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0051 - acc: 0.9991 - val_loss: 1.0428 - val_acc: 0.8251

<keras.callbacks.History at 0x12d3ccb00>

(IMPLEMENTATION) Load the Model with the Best Validation Loss

### TODO: Load the model weights with the best validation loss.
resnet50_model.load_weights('saved_models/weights.best.ResNet50.hdf5')

(IMPLEMENTATION) Test the Model

Try out your model on the test dataset of dog images. Ensure that your test accuracy is greater than 60%.

### TODO: Calculate classification accuracy on the test dataset.
# get index of predicted dog breed for each image in test set
resnet50_predictions = [np.argmax(resnet50_model.predict(np.expand_dims(feature, axis=0))) for feature in test_resnet50]

# report test accuracy
test_accuracy = 100*np.sum(np.array(resnet50_predictions)==np.argmax(test_targets, axis=1))/len(resnet50_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)

Test accuracy: 80.1435%

(IMPLEMENTATION) Predict Dog Breed with the Model

Write a function that takes an image path as input and returns the dog breed (Affenpinscher, Afghan_hound, etc) that is predicted by your model.

Similar to the analogous function in Step 5, your function should have three steps:

1. Extract the bottleneck features corresponding to the chosen CNN model.
2. Supply the bottleneck features as input to the model to return the predicted vector. Note that the argmax of this prediction vector gives the index of the predicted dog breed.
3. Use the dog_names array defined in Step 0 of this notebook to return the corresponding breed.

The functions to extract the bottleneck features can be found in extract_bottleneck_features.py, and they have been imported in an earlier code cell. To obtain the bottleneck features corresponding to your chosen CNN architecture, you need to use the function

extract_{network}

where {network}, in the above filename, should be one of VGG19, Resnet50, InceptionV3, or Xception.

### TODO: Write a function that takes a path to an image as input
### and returns the dog breed that is predicted by the model.
def Resnet50_predict_breed(img_path):
    # extract bottleneck features
    bottleneck_feature = extract_Resnet50(path_to_tensor(img_path))
    # obtain predicted vector
    predicted_vector = resnet50_model.predict(bottleneck_feature)
    # return dog breed that is predicted by the model
    return dog_names[np.argmax(predicted_vector)]

---

Step 6: Write your Algorithm

Write an algorithm that accepts a file path to an image and first determines whether the image contains a human, dog, or neither. Then,

• if a dog is detected in the image, return the predicted breed.
• if a human is detected in the image, return the resembling dog breed.
• if neither is detected in the image, provide output that indicates an error.

You are welcome to write your own functions for detecting humans and dogs in images, but feel free to use the face_detector and dog_detector functions developed above. You are required to use your CNN from Step 5 to predict dog breed.

Some sample output for our algorithm is provided below, but feel free to design your own user experience!

(IMPLEMENTATION) Write your Algorithm

### TODO: Write your algorithm.
### Feel free to use as many code cells as needed.
def prediction_pipeline(img_path):
    result = ""
    if(dog_detector(img_path)):
        result += "This is a dog. Its breed is: "
    elif(face_detector(img_path)):
        result += "If this human had been a dog,\n its breed would have been: "
    else: return "Can not detect neither human face nor dog on the image."
    
    return result + Resnet50_predict_breed(img_path)

---

Step 7: Test Your Algorithm

In this section, you will take your new algorithm for a spin! What kind of dog does the algorithm think that you look like? If you have a dog, does it predict your dog's breed accurately? If you have a cat, does it mistakenly think that your cat is a dog?

(IMPLEMENTATION) Test Your Algorithm on Sample Images!

Test your algorithm at least six images on your computer. Feel free to use any images you like. Use at least two human and two dog images.

Question 6: Is the output better than you expected :) ? Or worse :( ? Provide at least three possible points of improvement for your algorithm.

Answer: Well, as opposed to the 80% accuracy of the model, the results are rougly what I've expected. In the test images of dogs, the model gave only one mistake, incorrectly classified mops as chinese shar pei, but all other dog breeds were classified correctly. I find this a pretty good result, as dog breed classifying is a pretty tough task and many of the breeds have a lot of features in common. Even a human eye may make a mistake, labelling some of them. Anyway, this approach for classification may be improved, at first glance I can name a few ways of improvements:

• Enlarge set of images with different breeds (via collecting more data and via augmenting existing data) for training
• Train a specific mmodel for breed classification from scratch (although training a really good model would take a while)
• Implement specific pipeline for preprocessing dog images (for highlighting specific features, like color/fur length/size/tail form/etc)
• I would also add another step of preprocessing to define region of interest on the image, as having e.g. grass or other objects on the image may result in incorrect features maps and incorrect results. Defining a region of interest with just a dog would lead to more correct results.
• I also think that adding a few more layers to the used model and playing with them a bit would help incresing the accuracy slightly

## TODO: Execute your algorithm from Step 6 on
## at least 6 images on your computer.
## Feel free to use as many code cells as needed.
sample_files = np.array(glob("sample_images/**"))

images = []
titles = []
for file in sample_files:
    images.append(image.load_img(file))
    titles.append(prediction_pipeline(file))

def show_images(images, cols, titles):
    n_images = len(images)    
    fig = plt.figure()
    for n, (image, title) in enumerate(zip(images, titles)):
        a = fig.add_subplot(cols, np.ceil(n_images/float(cols)), n + 1)
        plt.imshow(image)
        a.set_title(title, fontsize=30)
    fig.set_size_inches(np.array(fig.get_size_inches()) * n_images)
    plt.show()

show_images(images, 3, titles)